Link aggregation

Me again :slight_smile:

I'm trying to find a page describing the L2, L3 und L4 modes of dladm.

It's nice to read "hashed by ip header", but how should I use that?

On the file-server it's ok to have the six interfaces serving six clients each on it's own. But an rsync connection via switch between two machines still uses only one link, is that true?

I dont undersatnd the logic behind your question. Of course a single link is used. What are you actually asking?:confused:

To use two links TOGETHER for a connection to another client :wink:

thats called trunking. are you referring to that?

Would you mind telling me the difference between link aggregation and trunking? As I thought it's the same?

AFAIK trunking and link aggregation are the same for the solaris/opensolaris context

just be sure, that your particular NIC driver supports it, and take a closer look at link-aggregation articles, like this in particular

ok ok, I probably haven't expressed it clearly enough:

I have set up two interfaces with link aggregation and they work just fine!

But: When I start an rsync process or scp or whatever similar, it only uses one physical interface. So what I am looking for is a way, to distribute traffic from a single process, e.g. rsync, to both interfaces.

bonding and trunking are dependent on your type of interface they are using.
trunking would be like a qfe and link-aggr would be something like nxge.

they are pretty much the same thing.

anyway, lets see the output of ifconfig -a, dladm show-aggr and netstat -rn . finally, how are you determining that rsync or some other file transfer is only using a single interface? i have to ask if the second link in the bonded pair is up.

obelix:/# ifconfig -a
lo0: flags=2001000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4,VIRTUAL> mtu 8232 index 1
inet 127.0.0.1 netmask ff000000
aggr1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
inet 192.168.17.8 netmask ffff0000 broadcast 192.168.255.255
ether 0:30:48:8a:c6:ee
obelix:/# dladm show-aggr
Schlüssel: 1 (0x0001) Richtlinie: L4 Adresse: 0:30:48:8a:c6:ee (Auto)
Gerät Adresse Geschwindigkeit Duplex Link Status
e1000g0 0:30:48:8a:c6:ee 1000 Mbps full up attached
e1000g1 0:30:48:8a:c6:ef 1000 Mbps full up attached
obelix:/# netstat -rn

Routing Table: IPv4
Destination Gateway Flags Ref Use Interface
-------------------- -------------------- ----- ----- ---------- ---------
default 192.168.17.1 UG 1 0
192.168.0.0 192.168.17.8 U 1 11 aggr1
224.0.0.0 192.168.17.8 U 1 0 aggr1
127.0.0.1 127.0.0.1 UH 4 76 lo0
obelix:/#

Yes, they are up. I ran an rsync for several hours and checked the statistics for both interfaces:

obelix:/# !375
dladm show-dev --statistics
ipackets rbytes ierrors opackets obytes oerrors
e1000g0 783204 55453484 0 20321 4288909 0
e1000g1 68599 8120481 0 7719021 10998514776 0

See the difference?